Agenda
Non-Differentility test: conditional data simulation
Pregnancy: predictive model
BRAHMS: PPV confidence interval
Non-Differentility test
### starting parameters
N <- 200000
prop_exp <- 0.2
pi_ne <- 0.05
risk <- 2
pi_e <- pi_ne*risk
### validation indices: SE_AUB
SE_AUB <- 0.90
### validation indices: E=1 & Y=1
SE_A_e <- 0.5
SE_B_given_A_e <- 0.8
SE_A_int_B_e <- SE_A_e * SE_B_given_A_e
SE_B_e <- SE_AUB - SE_A_e + SE_A_int_B_e
SE_A_given_B_e <- SE_A_int_B_e / SE_B_e
SE_B_given_not_A_e <- (SE_B_e - SE_A_int_B_e) / (1 - SE_A_e)
### validation indices: E=0 & Y=1
SE_A_ne <- 0.5
SE_B_given_A_ne <- 0.8
SE_A_int_B_ne <- SE_A_ne * SE_B_given_A_ne
SE_B_ne <- SE_AUB - SE_A_ne + SE_A_int_B_ne
SE_A_given_B_ne <- SE_A_int_B_ne / SE_B_ne
SE_B_given_not_A_ne <- (SE_B_ne - SE_A_int_B_ne) / (1 - SE_A_ne)
### validation indices: E=1 & Y=0
SP_A_e <- 0.99
SP_B_e <- 0.90
### validation indices: E=0 & Y=0
SP_A_ne <- 0.99
SP_B_ne <- 0.90
### Bootstrap parameters
na = 250 # A sample size
nb = 250 # B sample size
nsam = 1000 # Monte Carlo rep
nboot = 500 # Bootstrap sample size